-derived alpha control by inhibiting mRNA

Matthew Brooka,1, Gareth H. Tomlinsonb,1, Katherine Milesb,1, Richard W. P. Smitha, Adriano G. Rossib, Pieter S. Hiemstrac, Emily F. A. van ’t Woutc, Jonathan L. E. Deand, Nicola K. Graya, Wuyuan Lue,f, and Mohini Grayb,2

aMedical Research Council (MRC) Centre for Reproductive Health, Queen’s Medical Research Institute, University of Edinburgh, Edinburgh EH16 4TJ, Scotland; bMRC Centre for Inflammation Research, Queen’s Medical Research Institute, University of Edinburgh, Edinburgh EH16 4TJ, Scotland; cDepartment of Pulmonology, Leiden University Medical Center, 2300 RC, Leiden, The Netherlands; dKennedy Institute of Rheumatology, University of Oxford, Oxford OX3 7FY, United Kingdom; eInstitute of Human Virology, University of Maryland School of Medicine, Baltimore, MD 21201; and fDepartment of Biochemistry and Molecular , University of Maryland School of Medicine, Baltimore, MD 21201

Edited by Lionel B. Ivashkiv, Hospital for Special Surgery, New York, NY, and accepted by the Editorial Board February 29, 2016 (received for review February 9, 2016) are the first and most numerous cells to arrive at the within the azurophil granules of neutrophils, of which HNP1 is the site of an inflammatory insult and are among the first to die. We most abundant (6–9). They share a similar triple-stranded β-sheet previously reported that alpha defensins, released from apoptotic structure, which is critically held together by three intramolecular human neutrophils, augmented the antimicrobial capacity of disulphide bridges. Once the azurophil granules fuse with while also inhibiting the biosynthesis of proinflam- phagosomes, they release high concentrations of α-defensins close matory . In vivo, alpha administration protected to the surface, where their amphipathic nature allows mice from inflammation, induced by thioglychollate-induced peri- them to rapidly gain entry to the cell’s membrane (10). The per- tonitis or following infection with Salmonella enterica serovar meabilization of membranes by α-defensins is believed to be crucial Typhimurium. We have now dissected the antiinflammatory mech- for their ability to kill microbes and host cells, elicited by mem- anism of action of the most abundant neutrophil , brane disruption and leakage of cellular contents (9, 11). Impor- Human Neutrophil 1 (HNP1). Herein we show that HNP1 tantly, however, α-defensins only kill proliferating Escherichia coli enters macrophages and inhibits translation without in- and a simple model of “death by pore formation” is inadequate to ducing the unfolded-protein response or affecting mRNA stability. explain all their antibacterial properties (12). They have also been In a cell-free in vitro translation system, HNP1 powerfully inhibited noted to inhibit bulk bacterial protein synthesis in E. coli, although both cap-dependent and cap-independent mRNA translation while this is thought to be a consequence of membrane disruption and is maintaining mRNA polysomal association. This is, to our knowl- temporally associated with cell death (11, 12). Additionally, fol- edge, the first demonstration of a peptide released from one cell lowing HIV-1 infection, α-defensins play a crucial role in inhibiting type (neutrophils) directly regulating mRNA translation in another their life cycle (13, 14), suggesting that they have at their disposal a (macrophages). By preventing protein translation, HNP1 functions number of different mechanisms to kill diverse (7, 15). as a “molecular brake” on macrophage-driven inflammation, en- suring both pathogen clearance and the resolution of inflamma- Significance tion with minimal bystander tissue damage.

macrophages | α-defensins | mRNA translation | inflammation | cytokines Neutrophils are the major effectors of acute inflammation responding to tissue injury or infection. The clearance of apo- ptotic neutrophils by inflammatory macrophages also provides eutrophils, via the release of key inflammatory mediators, a powerful proresolution signal. Apoptotic or necrotic neutro- Nconvey signals to practically all other immune cells, orches- phils also release abundant amounts of the antimicrobial pep- trating both the innate inflammatory and subsequent adaptive tides alpha defensins. In this report, we show that the most immune responses (1). Through the de novo generation of abundant of these , HNP1 (Human Neutrophil Peptide 1), mediators, they are also key players in the resolution of inflammation profoundly inhibits protein translation. It achieves this without (reviewed in ref. 2). Following neutrophil apoptosis, their subsequent affecting mRNA stability or preventing mRNA polysomal associ- uptake by human monocyte-derived macrophages (HMDMs) in- ation. This is, to our knowledge, the first demonstration of a duces complex phenotypic changes, including the release of the im- peptide released from one cell, a leukocyte, entering and di- munosuppressive cytokines IL-10 and TGF-β (reviewedinref.3).We rectly modulating the translatome of another cell. It alludes to previously reported that the human α-defen- a previously unidentified mechanism, driven by dying neu- sins [which are released following apoptosis, necrosis, or NET-osis trophils, that ensures the timely resolution of macrophage- (4) of neutrophils] also inhibited the secretion of multiple cytokines driven inflammation, without compromising antimicrobial fromactivatedHMDMsforupto72h,withfullrecoverythereafter function. and no effect on cell viability (5). In vivo, in mice, neutrophil derived α-defensins, given at the time of inducing peritonitis, led to a di- Author contributions: M.B., N.K.G., W.L., and M.G. designed research; M.B., G.H.T., K.M., R.W.P.S., and E.F.A.v.t.W. performed research; A.G.R., P.S.H., J.L.E.D., and W.L. contributed minished inflammatory exudate (5). In addition, mice infected with new reagents/analytic tools; M.B., G.H.T., N.K.G., and M.G. analyzed data; M.B. and M.G. pathogenic Salmonella enterica serovar Typhimurium showed a re- wrote the paper. α duced bacterial load and serum TNF levels upon administration of The authors declare no conflict of interest. α α exogenous -defensin. Hence, neutrophil-derived -defensins were This article is a PNAS Direct Submission. L.B.I. is a guest editor invited by the Editorial able to affect profound changes in the inflammatory environment Board. while also serving as effective antimicrobial peptides. 1M.B., G.H.T., and K.M. contributed equally to this work. – Alpha defensins are small (3 4 kDa) cationic peptides that form 2To whom correspondence should be addressed. Email: [email protected]. part of a larger family of defensins (that also includes beta and This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. theta peptides). Four structurally related peptides (HNP1–4) exist 1073/pnas.1601831113/-/DCSupplemental.

4350–4355 | PNAS | April 19, 2016 | vol. 113 | no. 16 www.pnas.org/cgi/doi/10.1073/pnas.1601831113 Downloaded by guest on September 27, 2021 A BCD 12 Fig. 1. HNP1 inhibits bulk protein synthesis, which 3 IL6 Melle *** 10 is dependent on HNP1 tertiary structure. (A and B) IL1 HNP1-treated HMDMs were stimulated with the 8 HNP1 ** 2 * TLR7 ligand R848 (1 μg/mL) (A)orwith3μg/mL g/ml *** g/ml W26A n **** n 5 CD40L + 5 ng/mL IFNγ (B) for 18 h. TNFα (A) and IL-6 4 **** 1 * LHNP1 β * ** and IL-1 (B) were assayed by ELISA. (C and D) TNF ** TNF HNP1 ** *** R848 HMDMs were stimulated as for A and treated with

g/ml cytokine Melle 0 ** 12.5 μg/mL of HNP1 or the mutant peptides LHNP, 0 400 800 0 5 10 15 0 5 10 20 50 0 5 10 15 W26A, or Melle at the same (C) or variable con- HNP g/ml HNP g/ml TNF g/ml HNP g/ml centrations (D). TNFα assayed by ELISA after 18 h. EFResults are representative of five independent 4hr 18hr experiments. One-way ANOVA with Dunnett’s 1.5 Cellular Secreted Cellular Secreted ** multiple comparison tests; **P < 0.01, *P < 0.05. CD40/IFN --++--++ --++--++ (E) Methionine-starved HMDMs were then cultured ** * with 10 μCi/mL [35S]methionine ± activation (with HNP1 -+-+-+-+ -+-+-+-+ * 1.0 3 μg/mL CD40L and 5 ng/mL IFNγ) and ± addition of HNP1 (25 μg/mL) for 4 or 18 h. Secreted and in- 35S-Met- tracellular were resolved by SDS/PAGE. labelled Phosphorimages of radiolabeled cellular and se- proteins 0.5 creted protein gels show de novo protein synthesis. S-Met incorporation incorporation S-Met (F) De novo protein synthesis of 35S-Methionine– 35 labeled proteins following 18 h of culture, quanti- 0.0 fied by scintillation counting and normalized to un- C/IFN -- + ++- - + treated controls. n = 3. Error bars represent mean ± HNP1 - + - + --+ + SEM; **P < 0.01, *P < 0.05 (Tukey’s post hoc test Cellular Secreted following a one-way ANOVA).

In favor of this hypothesis is the observation that α-defensin di- secreted proteins in unstimulated HMDMs and robustly inhibited merization (which requires a tryptophan residue at position 26) is the labeling of secreted proteins in CD40L/IFNγ-stimulated vital for its ability to kill Staphylococcus aureus (16) but has little HMDMs, possibly reflecting the highly secretory phenotype of the effect on its ability to kill E. coli (17). stimulated macrophage. As expected, secreted TNFα was signifi- We wished to understand how α-defensins could simultaneously cantly reduced by HNP1 (Fig. S1A). However, the overall cellular function as an effective antimicrobial while also inducing protein levels were unchanged during the time course of the ex- profound changes in HMDM gene expression. We report here periment (Fig. S1B), consistent with a lack of increased global that HNP1 enters HMDMs, where it profoundly inhibits protein protein turnover and with maintenance of cell number and via-

translation in both resting and activated macrophages, without bility, as previously reported (5). Taken together, neutrophil- CELL BIOLOGY affecting mRNA stability or turnover. Instead it abrogates mRNA derived HNP1 profoundly inhibits global protein synthesis within translation without affecting mRNA polysomal association. the resting or activated macrophage.

Results Exogenous HNP1 Accumulates in the Macrophage. HNP1 gained HNP1 Inhibits the Synthesis of Proteins, Which Is Dependent on HNP1 entry to macrophages and was found within the membrane and Tertiary Structure. We have previously shown that although alpha cytoplasm. However, there was no clear colocalization of HNP1 defensins augmented the macrophage’s ability to kill intracellular (or the control peptide W26A) with the ER marker calreticulin Pseudomonas aeruginosa, these peptides simultaneously inhibited (Fig. 2A and Fig. S2 A and C) or with ribosomes (stained with the production of multiple cytokines (TNFα,IL-6,IL-8,andIL-1β) anti-Rps20; Fig. 2B and Fig. S2 B and D). Control experiments (5). HNP1 also inhibited TNFα biosynthesis from HMDMs stim- ulated with the toll-like receptor 7/8 (TLR7/8) agonist R848 (Fig. 1A). The biosynthesis of IL-6 and IL-1β induced via the T-cell surrogate stimulus CD40L/IFNγ was also reduced (Fig. 1B), con- A HNP1 Calreticulin Merge+ DAPI firming that disparate stimuli and multiple secreted proteins were susceptible to HNP1-mediated inhibition. The structure of HNP1 was crucial for its cytokine inhibitory potential. When the intra- molecular disulphide bonds that stabilize the triple-stranded beta- sheet structure of HNP1 were disrupted (linearized HNP1, L-HNP) 60 m or when dimerization was prevented by replacing the tryptophan residue at position 26 with the nonpolar (W26A) B HNP1 Rps20 Merge+ DAPI (16), a complete loss of cytokine inhibitory potential was seen (Fig. 1C and ref. 5). In contrast N-methylation of Ile20 (Melle), which also prevents dimerization, had a minimal effect on the ability of HNP1 to inhibit R848-induced TNFα production by HMDMs (Fig. 1 C and D). To test if HNP1 might inhibit protein synthesis per se, stimu- lated HMDMs were labeled with [35S]methionine in the presence of HNP1. [35S]methionine incorporation into proteins within cel- Fig. 2. HNP1 enters HMDMs. Confocal microscopy images of HNP1-treated HMDMs before visualization of anti-HNP1 (green) and DAPI (blue) seen on lular lysates (i.e., cellular proteins) and the culture media (i.e., the merged images. In addition, red secondary staining indicates calreticulin secreted proteins) was visualized (Fig. 1E) and quantified fol- (specific for the ER) in A and the ribosomal-associated protein Rps20 in B. lowing 18 h of culture (Fig. 1F). Strikingly, HNP1 treatment sig- Representative images are from one of six independent experiments. (White 35 nificantly reduced the quantity of both S-labeled cellular and scale bars, 60 μm.)

Brook et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4351 Downloaded by guest on September 27, 2021 ABD

CE

Fig. 3. HNP1 binds to mRNA but does not affect mRNA stability. (A) EMSA. Shown are the poly(C)25 RNA oligonucleotide probe (10 pmoles) incubated with molar ratios of HNP1 or W26A and RNA:peptide complexes resolved by nondenaturing acrylamide gel electrophoresis. Asterisk, free poly(C) probe; arrow-

head, nonspecific complex. Error bars represent mean ± SD. (B) Binding of HNP1 and W26A to poly(C)25 RNA relative to total input RNA (where the relative amount of free probe is given in arbitrary units). (C) RNA was extracted from CD40L/IFNγ-stimulated HMDMs, and mRNA of TNF-α, IL-10, TTP, and Cox-2 was quantified by quantitative real time-PCR (qRT-PCR) and expressed as the ratio of mRNA from treated to untreated HMDMs. (D) Supernatants were collected for the first 10 h from cells treated as in C and TNFα protein assayed by ELISA. (E) TNFα mRNA levels were quantified from HMDMs that had been treated with 12.5 μg/mL of HNP1 or W26A and then stimulated with R848 (1 μg/mL) for 1 h before adding actinomycin D (5 μg/mL). TNFα is expressed relative to T = 0 min. Error bars are mean ± SEM for each time point, and a line represents a nonlinear two-phase decay fit with R2 values of 0.8667 and 0.8351 for W26A and HNP1, respectively. Results are derived from three separate experiments. A–D represent experiments repeated three times. (A and B) Tukey’s post hoc test following a one-way ANOVA; ****P < 0.0001, *P < 0.03. (C and D) Tukey’s post hoc test following a two-way ANOVA. n.s., not significant; P = 0.094.

also showed no nonspecific staining or cross-reactivity between before the addition of actinomycin D to arrest further transcription. HNP1 and the ER or ribosomal secondary antibodies (Fig. S3). The decay rate of TNF-α mRNA was not significantly modulated in HNP1 versus W26A-treated HMDMs over a further 1-h time HNP1 Binds Nonspecifically to RNA but Does Not Alter mRNA course (Fig. 3E). As TNF-α mRNA stability is mediated in part by Transcription or Stability. As HNP1 enters the macrophage, it may, the zinc-finger protein TTP, which binds AU-rich sequences, we by reason of its positive charge and amphipathic nature (10, 18), bind also assessed TNF-α protein secretion from activated mouse bone to mRNA, so altering its turnover and inhibiting protein synthesis. marrow-derived macrophages (BMDMs) isolated from TTP- − − This was tested using electrophoretic mobility shift assays (EMSAs) deficient (TTP / ) mice or wild-type littermate controls. Again, with 25-mer homopolymeric RNA oligonucleotides. In contrast to HNP1 (but not L-HNP1) was still able to significantly inhibit the − − W26A, HNP1 showed concentration-dependent shifts of poly(C) secretion of TNF-α from TTP / BMDMs (Fig. S4G). Taken (Fig. 3 A and B), poly(A) (Fig. S4 A and B), and poly(U) RNA (Fig. together, these data show that HNP1 can bind to RNA, likely in S4 C and D), which were observed in both the presence or absence + a sequence-independent manner, but does not affect mRNA of Mg2 (Fig. S4E), a cation often required for nucleic acid binding stability or turnover. by proteins. An antibody supershift EMSA also confirmed that HNP1 could bind to mRNA [coding for the firefly luciferase (fLuc) HNP1 Does Not Induce ER Stress. We have previously shown that or β-galactosidase (β-gal) reporters] (Fig. S4F). HNP1 does not inhibit the exocytosis of TNFα from HMDMs (5). To ask if HNP1 affected mRNA transcription, we quantified the We also wished to confirm that it did not prevent protein synthesis steady-state mRNA levels generated by CD40L/IFNγ-stimulated by inducing the unfolded protein response (UPR) (reviewed in ref. HMDMs. The mRNA levels of TNFα,IL-10,cyclooxygenase 19). In contrast to the positive control thapsigargin (TG), we did (Cox2), and tristetraprolin (TTP) were unaffected by HNP1 not detect an increase in the synthesis of glucose-regulated protein treatment of HMDMs over a 24-h time course (Fig. 3C), despite a 78 (Grp78), X-box-binding protein (XBP1), or CCAAT/enhancer- clear reduction in TNFα protein production (Fig. 3D). To assess binding protein homologous protein (CHOP) in HNP1-treated mRNA decay, HNP1- or W26A-treated HMDMs were stimulated and -stimulated HMDMs (Fig. 4), despite a clear inhibition of (with R848) for 1 h, resulting in maximal TNF-α mRNA levels, R848-induced TNFα production at 6 and 24 h (Fig. S5A). Hence

4352 | www.pnas.org/cgi/doi/10.1073/pnas.1601831113 Brook et al. Downloaded by guest on September 27, 2021 absence of HNP1. Despite using a concentration of HNP1 that BiP 25 CHOP XBP-1 190 190 profoundly inhibited reporter protein synthesis (Fig. 1E), we 140 20 140 90 90 observed no change in the polysomal profile (Fig. 6A) or the 40 15 40 distribution of m7G-fLuc mRNA across the polysomal region of 10 10 10 the density gradient (fractions 4–10) (Fig. 6B). In contrast, the BiP mRNA mRNA BiP presence of EDTA resulted in polysomal dissociation and de- 5 mRNA CHOP 5 XBP-1 mRNA 5 pletion of the reporter mRNA from the fractions containing 0 0 0 0 6 12 18 24 0 6 12 18 24 0 6 12 18 24 translating mRNA (Fig. 6B and Fig. S5B). We also wished to Hours Hours Hours confirm if a similar mode of action was seen in HMDMs that had been treated with HNP1 or vehicle control (for 18 h). HMDMs Fig. 4. HNP1 does not cause ER stress. R848 (1 μg/mL) stimulated (filled di- so treated were then stimulated with R848 for 2 h to up-regulate amond) HMDMs with either 12.5 μg/mL HNP1 (filled inverted triangle), L-HNP1 the synthesis of TNFα. Again, the bulk polysome profile for (dot), or 1 μM TG (filled square). Macrophage mRNAs for CHOP, spliced XBP1, HNP1-treated HMDMs was similar to that of control-stimulated and BiP were quantified by qRT-PCR and expressed relative to the same mRNAs cells (Fig. 6C). Importantly, the polysomal association of TNFα in untreated control HMDMs. Hours represent time following stimulation. n = 3. Error bars represent mean ± SD. mRNA in untreated or HNP1-treated stimulated HMDMs was

the profound inhibition of protein synthesis by HNP1 was not the result of an induced UPR. A BC *** 0 1.6 10 HNP1 Does Not Block Translation Initiation. To ask if HNP1 affected 1.2 *** translation directly and to avoid the confounding effects of mRNA 1.2 10-1 transcription, processing, or nuclear export, we used the cell-free 0.8 rabbit reticulocyte lysate (RRL) in vitro translation system. 0.8 10-2 Translation of the canonical fLuc reporter mRNA was profoundly 0.4 inhibited in the presence of HNP1 but not by the mutant control 0.4 10-3 peptides L-HNP or W26A (Fig. 5A). As with TNFα mRNA, HNP1 Luciferase activity Luciferase activity did not destabilize the reporter mRNA because input mRNA 0.0 levels Luciferase mRNA 0.0 10-4 levels were maintained (Fig. 5B). The IC50 value for this effect was Veh 7 14 0 5 10 15 ∼1.6 μM(or5.5μg/mL) (Fig. 5C), a concentration that signifi- HNP1 M HNP1 W26A HNP1 µM cantly reduces the production of proinflammatory cytokines from Vehicle LHNP1 stimulated HMDMs in vitro (Fig. 1). * Eukaryotic mRNA has a 5′ monomethylated cap structure (m7G) DEF that is crucial for canonical translation initiation, the rate-limiting 1.2 100 * and primary node of translation regulation (reviewed in ref. 20). To 1.0 CELL BIOLOGY interrogate the role of translation initiation in HNP1-mediated in- hibition, we used reporter mRNAs that contained a viral internal 0.8 ribosome entry site (IRES) in their 5′ untranslated regions (5′ 10

-gal activity 0.5 UTRs), bypassing some or all of the eukaryotic translation initiation 0.4 factor (eIF) requirements and initiating translation cap-independently **** RLuc activity (reviewed in ref. 21). The Classical Swine Fever (CSFV) Relative mRNA IRES mRNA reporter initiates translation independently of the 0.0 1 0.0 majority of eIFs but is dependent on the ternary complex (eIF2, Veh HNP1 510 HNP1 W26A GTP, and tRNAi), whereas the Cricket Paralysis Virus (CrPV) Vehicle LHNP1 IRES allows the direct assembly of the 80S ribosome at the start Fig. 5. HNP1 inhibits protein synthesis downstream of translation initiation. codon, bypassing all canonical initiation factor requirements (22). 7 Remarkably, despite their diverse mechanisms of translation ini- (A)1ngmG-fLuc-A0 reporter mRNA, translated in vitro using the RRL with 25 μg/mL [7.3μM] HNP1, LHNP1, W26A, or vehicle control (0.01% acetic acid). tiation, HNP1 was also able to prevent the synthesis of both the β Translational output was quantified as relative fLuc activity (normalized to CSFV-driven translation of -Gal (Fig. 5D) and the CrPV-driven vehicle control-treated samples). Error bars represent mean ± SEM (n = 3). 7 translation of Renilla luciferase (RLuc) (Fig. 5E). As HNP1 is able (B) Similar to A but relative m G-luciferase-A0 reporter mRNA levels were to prevent the translation of mRNAs using diverse mechanisms of quantified by qRT-PCR. Black bars represent pretranslation levels and white translation initiation, it is most likely that it is acting downstream of bars the posttranslation levels. Shown are the results of three experiments. 7 this point. To confirm this empirically, ribosomal recruitment onto a (C) Similar to A, with 400 pg m G-fLuc-A0 reporter mRNA translated in the radiolabeled m7G-capped fLuc reporter mRNA was quantified in presence of increasing concentrations of HNP1. The IC50 (shown by the dotted line) is 1.6 ± 0.02 μM. Mean ± SEM from two independent experiments. the presence of cycloheximide to halt the 80S ribosome at the start (D) 1 ng CSFV IRES-β-gal-A0 reporter mRNA was in vitro-translated as for A. codon, preventing translation elongation. Although HNP1 weakly Values were plotted relative to vehicle control (n = 3). Error bars represent inhibited translation initiation at 5 min following mRNA addition, mean ± SEM (n = 3). For A and D,***P < 0.001, *P < 0.05 (analyzed by Tukey’s by 10 min similar maximal 80S recruitment to that seen in vehicle multiple comparison post hoc test following one-way ANOVA). (E) 1 ng CrPV β < control-treated extracts was observed (Fig. 5F), indicating only a IRES- -gal-A0 reporter mRNA translated as in A.****P 0.0001, analyzed by small reduction in the rate of 80S recruitment in the presence of unpaired t test. Values are plotted relative to vehicle control. (F) RRL was pretreated with 150 μg/mL cycloheximide and either 25 μg/mL HNP1 or vehicle HNP1 and supporting the conclusion that HNP1 predominantly 32 7 inhibits mRNA translation postinitiation. control. We then added 1 ng P-labeled m G-fLuc-A0 reporter mRNA for the indicated times (shown in minutes) before 15–30% sucrose density gradient fractionation. The graph depicts the relative amounts of mRNA sedimenting HNP1 Does Not Affect Ribosomal Association with mRNA. Finally to with initiating ribosomes, normalized to the amount recruited at 5 min in ask if luciferase mRNA was maintained on polysomes despite its vehicle control-treated RRL. Black bars represent the control group, and gray significantly reduced translation, we assessed the steady-state bars represent the HNP1-treated group. Error bars represent mean ± SEM (n = ribosomal association of m7G-fLuc mRNA in the presence or 3). *P < 0.05 (unpaired t test).

Brook et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4353 Downloaded by guest on September 27, 2021 AC 245nm 245nm Abs Abs

Fig. 6. HNP1 has no effect on polysome profile. (A) RRL pretreated with 25 μg/mL HNP1 or vehicle control. 32 7 1 2345678910 1 2 345678910 Shown is 2 ng P-labeled m G-fLuc-A0 reporter mRNA translated for 30 min before addition of 150 μg/mL cy- 10% Sucrose 50% Fraction cloheximide or 25 mM EDTA and 10–50% sucrose density gradient fractionation. Solid black line, vehicle control- BDtreated; broken black line, HNP1-treated; dotted gray 2.0 1.0 line, EDTA-treated. (B) Relative reporter mRNA content of gradient fractions expressed as a percentage of the total input mRNA. Solid black line with squares, vehicle control- 0.8 1.5 treated; broken black line with triangles, HNP1-treated; dotted gray line with filled circles, EDTA-treated. 0.6 (C) HMDMs treated with 25 μg/mL HNP1 or vehicle con- trol before R848 stimulation for 2 h. We added 150 μg/ 1.0 mL cycloheximide for 10 min before lysis and 10–50%

0.4 sucrose density gradient fractionation. An Abs254 nm trace was used to determine sedimentation of 80S ribosomes and polysomes. Solid black line, vehicle control-treated; 0.5 0.2 TNF dotted line, HNP1-treated. (D)TNFα mRNA content of gradient fractions expressed relative to maximal TNFα – Relative fLuc mRNA 0.0 mRNA detected in fractions 3 10 (43S/60S to polysomal). 2345 678 910 3 45678910 Solid black line, vehicle control-treated; broken black line, HNP1-treated. Error bars represent mean ± SD (n = Fraction 4); paired t test, no significant differences detected.

not significantly altered (Fig. 6D), despite the significant inhibition Previous reports point to several RNA-binding proteins that require of TNFα protein synthesis (Fig. S5C). These data confirm that a net positive charge and arginine side chains (18). Alpha defensins although HNP1 profoundly alters protein translation at a point also possess four positively charged arginines, which might allow it to after translation initiation, it does not prevent mRNA polysomal interact with RNA (Fig. 3). These side chains are important for its association. function, as the substitution of these amino acids for similarly charged lysine significantly reduces its bactericidal activity (17, 26) Discussion (reviewed in ref. 10). Considering the ability of HNP1 to kill a di- Cells of the have developed tightly regulated sys- verse array of bacterial and viral pathogens, it will be of interest to tems to ensure the timely resolution of inflammation. The control determine whether HNP1 can similarly prevent prokaryotic protein ofmRNAtranslationisemergingasamajormechanismthat translation. regulates the levels of proteins within leukocytes (reviewed in refs. Because HNP1 binds nonspecifically to RNA, we asked if it could 23, 24). We have now identified a previously unidentified mecha- inhibit translation by modulating ribosome engagement with mRNA. nism in which the most abundant neutrophil α-defensin, HNP1, However, both reporter and cellular mRNAs remained poly- which is readily released as these cells die (5), inhibits bulk protein some-associated (Figs. 5 and 6), and the polysomal distribution translation within macrophages. Although the characteristic hy- of these mRNAs was similar in control and HNP1-treated RRL drophobic, amphipathic nature of α-defensins allows them to and HMDMs. Translational repression could be occurring via partition into the membrane lipid layer (25), it also ensures ready either elongation and/or termination (27), and we would spec- access to the cell’s interior. Confocal imaging showed that HNP1 ulate that HNP1 prevents translation elongation (22), which has entered macrophages (Fig. 2) without inducing a UPR (Fig. 4) or recently been established as a major control point for protein affecting mRNA stability (Fig. 3). To our knowledge, this is the synthesis (28). first description of an eobiotic peptide released by one cell pro- Previous studies also allude to the greater importance of protein foundly affecting the translational capacity of another, in the ab- synthesis rate over degradation rate in determining overall protein sence of a requirement for de novo transcription and without levels (29, 30). However, the lack of a significant change in overall compromising antimicrobial function. HMDM cellular protein level (Fig. S1B) argues against an HNP1- HNP1 was able to inhibit translation initiation via diverse mech- mediated increase in nonspecific cellular protein degradation. anisms. Both canonical cap-dependent (Fig. 5) and noncanonical, Further, HNP1 profoundly inhibits reporter protein synthesis in cap-independent translation (driven by either a CSFV or CrPV cell-free assays in which protein turnover pathways are funda- IRES) were profoundly inhibited in vitro. However, the small in- mentally compromised and HNP1 itself has no known protease hibitory effect of HNP1 on translation initiation (Fig. 5F) was in- activity. Taken altogether, we believe these data indicate that sufficient to explain the magnitude of the effects seen in vitro and HNP1 affects de novo protein synthesis. within macrophages. Rather, the dramatic inhibition of CrPV IRES- The tertiary structure of monomeric HNP1 is also clearly im- driven translation, which dispenses with the initiation event, impli- portant for translational inhibition, as highlighted by the loss of cates an HNP1-mediated inhibition downstream of translation ini- efficacy observed for L-HNP1 or W26A (Fig. 1C). However, the tiation. HNP1 could inhibit translation by binding nonspecifically to N-methylation of HNP1 Ile-20 (Melle), which prevents dimerization, mRNA, or equally it could sequester factors essential for translation, does not alter the ability of Melle to inhibit TNF-α production, such as tRNA or ribosomal protein and/or rRNA components. confirming that HNP1 dimerization is not required to inhibit

4354 | www.pnas.org/cgi/doi/10.1073/pnas.1601831113 Brook et al. Downloaded by guest on September 27, 2021 macrophage protein translation (Fig. 1D). The concentration of Materials and Methods. Briefly, synthetic HNP1 and mutant derivatives were HNP1–3 in the synovial fluid of patients with rheumatoid ar- prepared by solid-phase synthesis as previously described (31). Template thritis is between 3 and 25 μg/mL, with an average of 12.4 μg/mL, plasmids pCSFV-lacZ (32) and pT7-Luc (33) for reporter mRNA transcription suggesting that the concentration reached in tissues is similar to were previously described, and pSL200-CrPV-RLuc, RLuc downstream of a CrPV that used in our assays (5). Our previous studies have shown that IRES, was a kind gift from Matthias Hentze, European Molecular Biology HMDMs fully recover their proinflammatory potential within Laboratory, Heidelberg, Germany. Healthy donor peripheral blood mono- 72 h following exposure to α-defensins. So although they clearly nuclear cells (PBMCs) were purified from whole blood as previously described μ μ disable the macrophage protein translation machinery, they do (5). Stimuli included 1 g/mL R848 (Invivogen), 3 g/mL CD40L (Peprotech), and γ not induce macrophage apoptosis (5). A previous study reported 5ng/mLIFN (Peprotech). Cytokines were quantified by sandwich ELISA (R&D that α-defensins reduced the release of IL-1β from activated Systems). For assessment of protein synthesis, HMDMs were incubated in β L-Methionine–free DMEM (MP Biomedicals) for 2 h at 37 °C, followed by monocytes, while not affecting the transcription of IL-1 mRNA μ 35 (28). Based on our findings, these observations can likely be 10 Ci/mL S-Methionine (Perkin-Elmer) and stimulation with CD40L and IFNγ and defensin peptides. In vitro transcription was assessed by m7G- or explained by the translation of pro–IL-1β being impaired. ApG-capped, nonadenylated, 32P-UTP–labeled or nonlabeled reporter mRNAs, In summary, we have uncovered that neutrophil α-defensins and the mRNAs were synthesized as previously described (34). In vitro trans- abrogate the bulk mRNA translation of proteins within HMDMs, lation was assessed using the nuclease-treated RRL in vitro translation kit without affecting mRNA transcription or stability. In this way they (Promega) according to the manufacturer’s recommendations. For EMSAs, prevent an excessive proinflammatory response that would create 10 pmoles 5′-Cy5–labeled 25-mer oligonucleotides (poly-Adenine, poly-Cytosine, its own collateral damage while still acting as powerful antimi- or poly-Uracil) (Eurogentec) were incubated with HNP1 or W26A peptide in crobial peptides. This is the first demonstration, to our knowledge, 10 μL binding and then resolved by electrophoresis. For immunocytochemistry, of an antimicrobial peptide that also has a translation-based HMDMs were grown on glass coverslips and stained with mouse monoclonal “ ” antiinflammatory role, acting as a molecular brake. It opens the anti-human HNP1–3 antibody together with polyclonal rabbit anti-human ri- way for developing similar peptide-based therapeutics that would act bosomal protein Rps20 (Abcam, dilution 1:250) or polyclonal rabbit anti-human as effective combined antiinflammatory and antimicrobial agents. calreticulin (Abcam, dilution 1:250). ER stress and the UPR and mRNA stability assay along with RNA quantitation and polysome analysis are fully explained in Materials and Methods SI Materials and Methods. All experiments on mice were covered by a project license granted by the Home Office under the Animal (Scientific Procedures) Act of 1986. Locally, this ACKNOWLEDGMENTS. This work was supported by Arthritis Research UK license was approved by the University of Edinburgh’s Ethical Review Com- Grant 18722 and Medical Research Council (MRC) Grant MR/J009555/1 (to mittee. All materials and the following protocols are fully described in SI M.G.) and by MRC Program Grant MR/J003069/1 (to N.K.G.).

1. Nathan C (2002) Points of control in inflammation. Nature 420(6917):846–852. 21. Firth AE, Brierley I (2012) Non-canonical translation in RNA . J Gen Virol 2. Serhan CN, Savill J (2005) Resolution of inflammation: The beginning programs the 93(Pt 7):1385–1409. end. Nat Immunol 6(12):1191–1197. 22. Fernández IS, Bai XC, Murshudov G, Scheres SH, Ramakrishnan V (2014) Initiation of 3. Savill J, Dransfield I, Gregory C, Haslett C (2002) A blast from the past: Clearance of translation by cricket paralysis virus IRES requires its translocation in the ribosome. – apoptotic cells regulates immune responses. Nat Rev Immunol 2(12):965 975. Cell 157(4):823–831. 4. Brinkmann V, et al. (2004) Neutrophil extracellular traps kill . Science 23. Piccirillo CA, Bjur E, Topisirovic I, Sonenberg N, Larsson O (2014) Translational control 303(5663):1532–1535. CELL BIOLOGY of immune responses: From transcripts to translatomes. Nat Immunol 15(6):503–511. 5. Miles K, et al. (2009) Dying and necrotic neutrophils are anti-inflammatory secondary 24. Carpenter S, Ricci EP, Mercier BC, Moore MJ, Fitzgerald KA (2014) Post-transcriptional to the release of alpha-defensins. J Immunol 183(3):2122–2132. regulation of gene expression in innate immunity. Nat Rev Immunol 14(6):361–376. 6. Ganz T, Lehrer RI (1997) Antimicrobial peptides of leukocytes. Curr Opin Hematol 4(1): 25. Brogden KA (2005) Antimicrobial peptides: Pore formers or metabolic inhibitors in 53–58. bacteria? Nat Rev Microbiol 3(3):238–250. 7. Lehrer RI (2004) Primate defensins. Nat Rev Microbiol 2(9):727–738. 26. de Leeuw E, Rajabi M, Zou G, Pazgier M, Lu W (2009) Selective arginines are impor- 8. Lehrer RI, Ganz T (2002) : A family of endogenous antimicrobial peptides. Curr Opin Hematol 9(1):18–22. tant for the antibacterial activity and host cell interaction of human alpha-defensin 5. – 9. Lehrer RI, Lichtenstein AK, Ganz T (1993) Defensins: Antimicrobial and cytotoxic FEBS Lett 583(15):2507 2512. peptides of mammalian cells. Annu Rev Immunol 11:105–128. 27. Richter JD, Coller J (2015) Pausing on polyribosomes: Make way for elongation in 10. Lehrer RI, Lu W (2012) α-Defensins in human innate immunity. Immunol Rev 245(1): translational control. Cell 163(2):292–300. 84–112. 28. Shi J, et al. (2007) A novel role for defensins in intestinal homeostasis: Regulation of 11. Kagan BL, Selsted ME, Ganz T, Lehrer RI (1990) Antimicrobial defensin peptides form IL-1beta secretion. J Immunol 179(2):1245–1253. voltage-dependent ion-permeable channels in planar lipid bilayer membranes. Proc 29. Schwanhäusser B, et al. (2011) Global quantification of mammalian gene expression Natl Acad Sci USA 87(1):210–214. control. Nature 473(7347):337–342. 12. Lehrer RI, et al. (1989) Interaction of human defensins with Escherichia coli. Mecha- 30. Kristensen AR, Gsponer J, Foster LJ (2013) Protein synthesis rate is the predominant nism of bactericidal activity. J Clin Invest 84(2):553–561. regulator of protein expression during differentiation. Mol Syst Biol 9:689. 13. Furci L, Sironi F, Tolazzi M, Vassena L, Lusso P (2007) Alpha-defensins block the early 31. Wu Z, Ericksen B, Tucker K, Lubkowski J, Lu W (2004) Synthesis and characterization steps of HIV-1 infection: Interference with the binding of gp120 to CD4. Blood 109(7): of human alpha-defensins 4-6. J Pept Res 64(3):118–125. – 2928 2935. 32. Smith RW, et al. (2011) DAZAP1, an RNA-binding protein required for development 14. Gallo SA, et al. (2006) Theta-defensins prevent HIV-1 Env-mediated fusion by binding and spermatogenesis, can regulate mRNA translation. RNA 17(7):1282–1295. – gp41 and blocking 6-helix bundle formation. J Biol Chem 281(27):18787 18792. 33. Gallie DR (1991) The cap and poly(A) tail function synergistically to regulate mRNA 15. Ganz T (2003) Defensins: Antimicrobial peptides of innate immunity. Nat Rev translational efficiency. Genes Dev 5(11):2108–2116. Immunol 3(9):710–720. 34. Gray NK (1998) Translational control by repressor proteins binding to the 5’UTR of 16. Wei G, et al. (2010) Trp-26 imparts functional versatility to human alpha-defensin mRNAs. Methods Mol Biol 77:379–397. HNP1. J Biol Chem 285(21):16275–16285. 35. Taylor GA, et al. (1996) A pathogenetic role for TNF alpha in the syndrome of ca- 17. Zou G, et al. (2007) Toward understanding the cationicity of defensins. Arg and Lys chexia, arthritis, and autoimmunity resulting from tristetraprolin (TTP) deficiency. versus their noncoded analogs. J Biol Chem 282(27):19653–19665. – 18. Tan R, Frankel AD (1995) Structural variety of arginine-rich RNA-binding peptides. Immunity 4(5):445 454. ’ Proc Natl Acad Sci USA 92(12):5282–5286. 36. van Schadewijk A, van t Wout EF, Stolk J, Hiemstra PS (2012) A quantitative method 19. Rutkowski DT, Kaufman RJ (2004) A trip to the ER: Coping with stress. Trends Cell Biol for detection of spliced X-box binding protein-1 (XBP1) mRNA as a measure of en- 14(1):20–28. doplasmic reticulum (ER) stress. Cell Stress Chaperones 17(2):275–279. 20. Hinnebusch AG (2014) The scanning mechanism of eukaryotic translation initiation. 37. Burgess HM, Gray NK (2012) An integrated model for the nucleo-cytoplasmic trans- Annu Rev Biochem 83:779–812. port of cytoplasmic poly(A)-binding proteins. Commun Integr Biol 5(3):243–247.

Brook et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4355 Downloaded by guest on September 27, 2021